Remembering readings: manually or by association?

[J44xm]

Good day.

Now, in my early kanji endeavors, I've been finding that I'm not sure what kind of mental process I should give priority to for remembering kanji readings. My two candidates are (1) trying to remember readings for each character individually (including distinguishing between onyumi and kunyomi) and (2) remembering words and sort of reverse engineering the readings like that (e.g., 「体重」 is "taijuu" and, thus, 「体」 can be read as 「たい」).

I'm not sure that I can do one very well, and since I'll be endeavoring to attain a larger vocabulary anyway, I'm leaning toward method two. (Remember, this is a discussion of priority, so it's not an either-or choice I'm talking about. I'm just seeking advice about which I should focus more on.)

Many thanks for any help.

[Bucko]

I learnt the English meanings of six hundred or so kanji, now I'm going through that list and creating word lists of the on-yomi pronounciations, then using JDICT (http://www.csse.monash.edu.au/~jwb/wwwjdic.html) to put together lists of sentences to remember them in context. For example, I just learnt a whole bunch of words using the kanji 火 with the on-yomi pronounciation か. I put the following words into sentences: 火口　（かこう）、火災（かさい）、火山（かざん）、火事（かじ）、大火事（おおかじ）、an d 火曜日（かようび）. Doing this really drills in the fact that 火 has the on-yomi pronounciation　か.

My plan is to go back through the list when I finish to learn the more difficult kun-yomi pronounciations.